\begin{table}[htbp]
\centering
\caption{Comparing impacts on fuel-collection time for all fuels and traditional fuels\label{app:fuel_collect_tab}}
\begin{threeparttable}
\begin{tabular}{l*{2}{c}}
\toprule
& (1) & (2) \\
& \multicolumn{2}{c}{Fuel-collection time (minutes per day)} \\ \cmidrule{2-3}
& All fuels & Traditional fuels \\
\midrule
\({POST}_1\)&       -19.0   &       -13.8   \\
            &      (18.5)   &      (19.0)   \\ \addlinespace
\({POST}_2\)&       -30.7** &       -44.4***\\
            &      (12.5)   &      (11.4)   \\ \addlinespace
\({TREATMENT}_j \times {POST}_1\)&        19.2   &        17.4   \\
            &      (25.4)   &      (24.8)   \\ \addlinespace
\({TREATMENT}_j \times {POST}_2\)&        12.8   &        15.4   \\
            &      (15.9)   &      (14.8)   \\ \addlinespace
\({NGO}_j \times {POST}_1\)&        53.3   &        52.0   \\
            &      (33.2)   &      (31.7)   \\ \addlinespace
\({NGO}_j \times {POST}_2\)&        48.5*  &        55.4** \\
            &      (29.0)   &      (25.1)   \\ \addlinespace
\({TREATMENT}_j \times {NGO}_j \times {POST}_1\)&       -95.1** &       -87.1** \\
            &      (40.5)   &      (38.9)   \\ \addlinespace
\({TREATMENT}_j \times {NGO}_j \times {POST}_2\)&       -60.0*  &       -55.5*  \\
            &      (32.2)   &      (28.4)   \\
\midrule
Baseline non-NGO control mean & 114 & 107 \\
Household fixed-effects & Yes & Yes \\
\(N\)       &       2,829   &       2,829   \\
Adjusted \(R^2\)&       0.028   &       0.031   \\
\bottomrule
\end{tabular}
\begin{tablenotes}
{\setlength\labelsep{0pt}
\footnotesize
\item \textit{Notes}. The outcome variable for fuel-collection time in column (1) is derived from self-reported data on time spent (per day, week, or month) collecting fuelwood, crop residue, leaves, dung, biomass pellets, kerosene, LPG, biogas, and---if relevant---any other fuel used by the household; the results in column (1) are identical to those presented in column (4) of Table \ref{t:impacts}. In column (2), fuel-collection time is restricted to only traditional fuels (fuelwood, crop residue, leaves, and dung). Standard errors (in parentheses) are clustered at the hamlet level. \sym{*} \(p<0.10\), \sym{**} \(p<0.05\), \sym{***} \(p<0.01\).}
\end{tablenotes}
\end{threeparttable}
\end{table}
